New C-C coupling Reactions Enabled by Main-group Organometallics

New C-C coupling Reactions Enabled by Main- group Organometallics

Kilian Colas

Academic dissertation for the Degree of Doctor of Philosophy in Organic Chemistry at Stockholm University to be publicly defended on Friday 16 November 2018 at 10.00 in Magnélisalen, Kemiska övningslaboratoriet, Svante Arrhenius väg 16B.

Abstract

The carbon-carbon bond has always been at the very core of chemical research. Strategies for the creation of C−C bonds are one of the keys to the construction game that organic chemists play with the building blocks provided by Nature, with the ultimate goal of producing useful molecular structures that will serve society as medicines, materials, imaging tools, catalysts, and ligands (to mention but a few). While very different in their structure, all of these molecules are often prepared by the same methods. However, efficiency could be improved with tailored chemical strategies that would serve an individual purpose. Ideally, these chemical manipulations should be efficient, selective, environmentally friendly and economic, in order to truly fulfill their final objective.

However, despite the ever-expanding rule-book of chemical reactions, target molecules of increasing complexity often face chemists with daunting challenges, whose success rely on multi-step synthetic sequences. There is therefore a permanent need for new, specific methods and strategies that are capable of seamlessly creating C−C bonds, evading the synthesis of difficult or expensive substrates. In this regard, common organometallic reagents display a unique behavior as carbon precursors, in particular as powerful nucleophiles. Reagents based on main-group elements such as lithium or magnesium have therefore played a central role in organic synthesis ever since their discovery. The challenge often lies in controlling their high reactivity, as well as their basic character. Tuning and taming these properties provides chemists with a wide range of unique strategies for the selective synthesis of countless molecular targets.

In the first part of this thesis, a scalable and stereoselective [3+3] homocoupling of imines in which two C−C bonds are formed in a single step is reported. This reaction relies on an unusual combination of visible-light irradiation and aluminum organometallics. This photochemical process enables the circumvention of the native [3+2] reactivity of these readily available starting materials, thus enabling rapid access to densely functionalized piperazines. Thanks to the congested environment they provide, these heterocyclic scaffolds can be used as ligands to prevent catalyst deactivation through oligomerization.

The next chapter presents a novel Pummerer-type redox-neutral coupling of sulfoxides and Grignard reagents. This reaction is enabled by a unique turbo-magnesium amide base, and allows the use of a wide range of carbon nucleophiles in intermolecular Pummerer C−C coupling for the streamlined preparation of thioethers. Given the central character of sulfur in organic chemistry, these compounds can then be converted to a variety of unrelated functional groups for the streamlined preparation of diverse building blocks.

In the final two chapters, the development of a method for the direct conversion of carboxylic acids to ketones with Grignard reagents is described. Using the above-mentioned combination of organometallics, a wide variety of carboxylic acids substrates and Grignard reagents can be coupled in a convenient, scalable and highly selective method that suppresses the need for activation and offers a straightforward approach to ketones from readily available starting materials.

Keywords: C-C coupling, organometallics, aluminum, magnesium, Grignard, turbo-Hauser-bases, piperazines, sulfur, ketones.

Stockholm 2018

http://urn.kb.se/resolve?urn=urn:nbn:se:su:diva-160658

ISBN 978-91-7797-418-5 ISBN 978-91-7797-419-2

Department of Organic Chemistry

Stockholm University, 106 91 Stockholm

NEW C-C COUPLING REACTIONS ENABLED BY MAIN-GROUP ORGANOMETALLICS

Kilian Colas

New C-C coupling Reactions Enabled by Main-group

Organometallics

Kilian Colas

©Kilian Colas, Stockholm University 2018 ISBN print 978-91-7797-418-5

ISBN PDF 978-91-7797-419-2

Printed in Sweden by Universitetsservice US-AB, Stockholm 2018 Distributor: Department of Organic Chemistry

"Ever tried. Ever failed.

 No matter.

 Try again. Fail again.

 Fail better."

Samuel Beckett

Abstract

The carbon-carbon bond has always been at the very core of chemical research. Strategies for the creation of C−C bonds are one of the keys to the construction game that organic chemists play with the building blocks provided by Nature, with the ultimate goal of producing useful molecular structures that will serve society as medicines, materials, imaging tools, catalysts, and ligands (to mention but a few). While very different in their structure, all of these molecules are often prepared by the same methods.

However, efficiency could be improved with tailored chemical strategies that would serve an individual purpose. Ideally, these chemical manipulations should be efficient, selective, environmentally friendly and economic, in order to truly fulfill their final objective.

However, despite the ever-expanding rule-book of chemical reactions, target molecules of increasing complexity often face chemists with daunting challenges, whose success rely on multi-step synthetic sequences. There is therefore a permanent need for new, specific methods and strategies that are capable of seamlessly creating C−C bonds, evading the synthesis of difficult or expensive substrates. In this regard, common organometallic reagents display a unique behavior as carbon precursors, in particular as powerful nucleophiles. Reagents based on main-group elements such as lithium or magnesium have therefore played a central role in organic synthesis ever since their discovery. The challenge often lies in controlling their high reactivity, as well as their basic character.

Tuning and taming these properties provides chemists with a wide range of unique strategies for the selective synthesis of countless molecular targets.

In the first part of this thesis, a scalable and stereoselective [3+3]

homocoupling of imines in which two C−C bonds are formed in a single step is reported. This reaction relies on an unusual combination of visible-light irradiation and aluminum organometallics. This photochemical process enables the circumvention of the native [3+2]

reactivity of these readily available starting materials, thus enabling rapid access to densely functionalized piperazines. Thanks to the congested

environment they provide, these heterocyclic scaffolds can be used as ligands to prevent catalyst deactivation through oligomerization.

The next chapter presents a novel Pummerer-type redox-neutral coupling of sulfoxides and Grignard reagents. This reaction is enabled by a unique turbo-magnesium amide base, and allows the use of a wide range of carbon nucleophiles in intermolecular Pummerer C−C coupling for the streamlined preparation of thioethers. Given the central character of sulfur in organic chemistry, these compounds can then be converted to a variety of unrelated functional groups for the streamlined preparation of diverse building blocks.

In the final two chapters, the development of a method for the direct conversion of carboxylic acids to ketones with Grignard reagents is described. Using the above-mentioned combination of organometallics, a wide variety of carboxylic acids substrates and Grignard reagents can be coupled in a convenient, scalable and highly selective method that suppresses the need for activation and offers a straightforward approach to ketones from readily available starting materials.

List of Publications

This report is based on the following papers/manuscripts, referred to in the text by Roman numerals I-IV.

I. Scalable Synthesis of Piperazines Enabled by Visible- Light Irradiation and Aluminum Organometallics

Samuel Suárez-Pantiga, Kilian Colas, Magnus J. Johansson and Abraham Mendoza*

Angew. Chem. Int. Ed. 2015, 54, 14094-14098

II. Intermolecular Pummerer Coupling with Carbon Nucleophiles in Non-Electrophilic Media

Kilian Colas, Raúl Martín-Montero and Abraham Mendoza*

Angew. Chem. Int. Ed. 2017, 56, 16042-16046

III. Iterative Synthesis of Pluripotent Thioethers through Controlled Redox-Fluctuation of Sulfur

Kilian Colas and Abraham Mendoza*

Synlett 2018, 29, 1329-1333

IV. Synthesis of Ketones from Carboxylic Acids Using Grignard Reagents and turbo-Hauser Bases

Kilian Colas. A. Catarina V. D. dos Santos and Abraham Mendoza*

manuscript in preparation

V. One-pot Synthesis of Ketones from Aliphatic Carboxylic Acids

Kilian Colas, A. Catarina V. D. dos Santos and Abraham Mendoza*

manuscript in preparation

Publications by the author not included as part of this report

Chemical Innovation through Ligand Total Synthesis Abraham Mendoza*, Kilian Colas, Samuel Suárez-Pantiga, Daniel Goetz, Magnus J. Johansson

Synlett 2016, 27, 1753-1759

Structure-Activity Relationship, Drug Metabolism and Pharmacokinetics Properties Optimization, and in Vivo Studies of New Brain Penetrant Triple T-Type Calcium Channel Blockers

Siegrist, R.; Pozzi, D.; Jacob, G.; Torrisi, C.; Colas, K.;

Braibant, B.; Mawet, J.; Pfeifer, T.; de Kanter, R.; Roch, C.;

Kessler, M.; Corminboeuf, O.; Bezençon, O.*

J. Med. Chem. 2016, 59, 10661-10675

Abbreviations

Abbreviations are used in agreement with the standards of the subject*.

Additional abbreviations used in this thesis are given below:

bpy bipyridine

cataCXium A di(1-adamantyl)-n-butylphosphine

CFL compact fluorescent light

DBH 1,3-dibromo-5,5-dimethylhydantoin

DCM dichloromethane

DIBAL-H diisobutylaluminum hydride

DIPA diisopropylamine

dipp 2,6-diisopropylphenyl

DOSY diffusion ordered spectroscopy

DPPF 1,1’-bis(diphenylphosphino)ferrocene

dtbbpy di-tert-butyl-bipyridine

equiv. equivalent(s)

LDA lithium diisopropylamide

LED light-emitting diode

pin pinacol

ppy phenylpyridine

py pyridine

r.t. room temperature

SET single-electron transfer

TIB tri-isopropylbenzoate

TMP 1,1,2,2-tetramethylpiperidine

TMS trimethylsilyl

TPA tris(2-pyridylmethyl)amine

*The ACS Style Guide, 3^rd Edition, American Chemical Society:

Washington, DC, 2006.

Contents

Abstract ... vii

List of Publications ... ix

Abbreviations ... xi

1. Introduction ... 1

1.1Organometallic compounds ... 1

1.2Organoaluminum reagents ... 2

1.3Organomagnesium reagents ... 4

1.3.1Grignard reagents ... 4

1.3.2turbo-Grignard reagents ... 7

1.4Magnesium amides ... 7

1.4.1Hauser bases ... 7

1.4.2turbo-Hauser bases ... 9

1.4.3Organomagnesium amides ... 10

1.5 Aims of the thesis ... 12

2. Scalable Synthesis of Piperazines Enabled by Visible-Light Irradiation and Aluminum Organometallics ... 13

2.1 Introduction ... 13

2.2 Aims of the project ... 15

2.3 Results and discussion... 16

2.3.1Reaction conditions ... 16

2.3.2Substrate scope ... 18

2.3.3Mechanistic studies... 20

2.3.4Application of the products in catalysis ... 21

2.4 Conclusion and outlook ... 23

3. Intermolecular Pummerer Coupling with Carbon Nucleophiles in Non-Electrophilic Media ... 24

3.1 Introduction ... 24

3.1.1Thioethers ... 24

3.1.2The Pummerer reaction ... 26

3.2 Aims of the project ... 29

3.3 Results and discussion... 29

3.3.1Reaction conditions ... 29

3.3.2Substrate scope and limitations ... 32

3.3.3 Iteration of the reaction and diversification of the products into various building blocks ... 38

3.3.4Preliminary mechanistic investigations. ... 41

3.4 Conclusion and outlook ... 43

4. Synthesis of Arylketones from Carboxylic Acids Using Grignard Reagents and turbo-Hauser Bases ... 44

4.1 Introduction ... 44

4.2 Aims of the project ... 47

4.3 Results and discussion... 48

4.3.1Reaction conditions ... 48

4.3.2Substrate scope and limitations ... 50

4.3.3Application to the coupling of Grignard reagents and carbon dioxide ... 53

4.3.4Mechanistic investigations ... 53

4.4 Conclusion and outlook ... 55

5. Synthesis of Ketones from Aliphatic Carboxylic Acids ... 57

5.1 Introduction ... 57

5.2 Aims of the project ... 59

5.3 Results and discussion... 59

5.3.1Reaction conditions ... 59

5.3.2Substrate scope and limitations ... 62

5.3.3Reaction conditions for primary acids ... 63

5.4 Conclusion and outlook ... 66

Concluding remarks ... 67

Populärvetenskaplig sammanfattning ... 68

Appendix A: Contribution List ... 70

Appendix B: Reprint permissions ... 71

Acknowledgements ... 72

References ... 74

1. Introduction

1.1 Organometallic compounds

Organometallics are compounds that contain a C‒M bond, where M represents a main-group metal, a transition metal or a metalloid element such as boron or silicon.^1a The polarity of the C‒M bond defines organometallic compounds as both powerful nucleophiles and strong bases (Scheme 1).

Depending on the metal used, they can be prepared through numerous methods such as direct metal insertion, halogen-metal exchange or directed metalation (see sections 1.2 and 1.3.1 for details).^1b-2

Scheme 1: Organometallics behave as nucleophiles and bases.

In general, the behavior of organometallic species greatly depends on the nature of the metal. For instance, compounds based on strongly electropositive metals are extremely reactive and less tolerant to existing functionalities. In contrast, organometallics based on less electropositive metals such as zinc are less reactive, but also more tolerant, while compounds based on boron or aluminum display unique reactivities due to their Lewis acidic p-orbital.^3a Organometallics are thus very versatile tools for organic synthesis, in particular for the construction of C‒C bonds (Scheme 2).^2a They readily add to electrophiles such as carbonyls or epoxides to provide alcohols, while some (based on copper or zinc, for instance) can instead partake in 1,4-additions with Michael acceptors.^1b-2a Several organometallics (based on boron, magnesium, zinc, tin, …) are also substrates of cross-coupling reactions.⁴ As bases, organometallics (lithium-

ortho-metalation reactions,⁶ thus providing an even greater variety of carbon-nucleophiles.^2a

Scheme 2: Organometallics can react as nucleophiles and bases.

The partially filled d-orbitals of transition-metal based organometallics give them a donor/acceptor character and allows them to engage in both σ- and π-bonding.1c, 2b, 7

These properties also explain why the reactivity of organometallic compounds of titanium (which has an empty d-orbital) and zinc (which has a filled d-orbital) resembles that of main-group organometallics.^2b

1.2 Organoaluminum reagents

Aluminum organometallics (organoalanes) are common reagents that can be prepared through reduction of organic halides, transmetallation, acid-base reactions or hydrometalation (Scheme 3).^3b The direct insertion of aluminum into a carbon‒halogen bond (Scheme 3, A) is desirable but only practical with alkyl-, allyl- or benzyl halides. This approach is mostly used for the preparation of trialkylaluminum reagents on industrial scales. As a matter of fact, AlMe3 is the organometallic reagent that is produced on the largest volume in the chemical industry, primarily to manufacture Ziegler-Natta polymerization catalysts.^3b When it comes to aryl-halides, the activation of

additives (such as AlCl3, LiCl or PbCl2).^3b Due to these shortcomings, arylaluminum compounds are commonly obtained from the corresponding arylmagnesium (or aryllithium) reagents by transmetallation (Scheme 3, B).

Alternatively, arylaluminum compounds can also be prepared by directed alumination from arenes using aluminum amide bases (Scheme 3, C).

Finally, hydroalumination of alkynes and alkenes provides vinyl- or alkyl- aluminum species upon treatment with DIBAL-H (Scheme 3, D).

Scheme 3: Strategies for the preparation of organoalanes.

The structure of organoalanes depends on the substituents on the metal center (Scheme 4).^3c Trialkylaluminums readily form dimeric species that are noteworthy for their pentavalent bridging carbons, an example of 3-centre-2- electron bonding. Likewise, the triphenylaluminum dimer is remarkable for its bridging phenyl rings. In contrast, monomeric compounds like Al(^tBu)3

occur when increasing the steric hindrance on the aluminum atom.

Scheme 4: Organoaluminum compounds can have dimeric or monomeric structures.

The vacant p-orbital and high oxophilicity of aluminum make these organometallics particularly Lewis acidic. In addition, organolaluminum reagents display a mild nucleophilicity. This dual reactivity has been exploited in numerous synthetic applications (Scheme 5).³ For instance, they

nucleophiles towards carbonyls,⁹ epoxides,¹⁰ imines¹¹ and Michael acceptors.¹² Organoaluminum reagents are also employed as reducing agents. For instance, DIBAL-H is known for its unique ability to prevent overreduction.¹³ Finally, organoalanes can also engage in more unusual applications, such as C-F activation¹⁴or cross-coupling reactions.¹⁵

Scheme 5: The Lewis acidity and nucleophilicity of organoalanes allows for a variety of synthetic applications.

1.3 Organomagnesium reagents

1.3.1 Grignard reagents

Organomagnesium compounds of the general formula RMgX are known as Grignard reagents. They were first reported by Victor Grignard in 1900, a discovery which was recognized by the 1912 Nobel prize.¹⁶ Several methods exist for their preparation (Scheme 6).^2a The direct oxidative addition of magnesium into a carbon−halogen bond is the most conventional approach (Scheme 6, A). The magnesium turnings must be activated, using catalytic amounts of I2,¹⁷ dibromoethane¹⁸ or DIBAL-H,¹⁹ or via mechanical²⁰ or flow techniques.²¹ The process is exothermic and often run in refluxing solvent.

Therefore, sensitive substrates require lower temperatures and the more reactive Rieke magnesium.²² However, significantly better functional group tolerance is currently achieved through halogen-magnesium exchange,

⁰ 2a, 23

Grignard reagents can be prepared by direct metalation of arenes with magnesium amide bases (Scheme 6, C), also called Hauser bases, which will be discussed in more detail later in this introduction (see section 1.4.1).

Scheme 6: Grignard reagents are prepared through various methods.

After more than a century of research, the precise structural aspects of Grignard reagents remain a notoriously mysterious matter.²⁴ While the simplified representation “RMgX” is widely used, it is well known that they form multimeric aggregates in solution. This is further complicated by disproportionation of heteroleptic RMgX to homoleptic diorganomagnesium R2Mg and magnesium halide MgX2, known as the Schlenk equilibrium (Scheme 7). Grignard reagents are therefore highly dynamic species

Scheme 7: The Schlenk equilibrium complicates the structural behavior of Grignard reagents.

Schlenk and other aggregation equilibria differ in solution and in the solid state, making research on the topic even more challenging. In 2016, detailed studies were published by the groups of Stalke and Koszinowski, exploiting a combination of mass-spectrometry, electrical conductivity measurements, DFT calculations and diffusion-ordered spectroscopy (DOSY) NMR analysis.²⁴ This study shows how the aggregation level and the position of the equilibrium are influenced by factors such as solvent, temperature, concentration and the nature of the organic ligand. In THF for instance, Grignard reagents are mostly found as monomers with significant amounts of homoleptic species. On the other hand, in dioxane solutions the equilibrium is fully shifted towards the diorganomagnesium species R2Mg due to the capacity of the solvent to sequester the magnesium halide MgX2.

Grignard reagents have been an indispensable part of the chemist arsenal ever since their discovery.²⁵ They efficiently add to carbonyl groups to provide alcohols,^{2a, 26} aldehydes,²⁷ ketones²⁸ and carboxylic acids²⁹ (Scheme 8). They have also found use in the construction of C−O,³⁰ C−N,³¹ C−S,³² C−Se,^32a C−P³³ or C−B³⁴ bonds, to name but a few. This wide scope of applications can be explained by the balanced reactivity-selectivity

analogues.¹ As such, they can generally be used at room temperature or higher with reasonably good chemoselectivities.^25b

Scheme 8: Grignard reagents react with a variety of electrophiles.

Grignard reagents also engage in asymmetric reactions with transition metal catalysis, for example in 1,4-additions to Michael acceptors (Scheme 9, A).³⁵ They are also known for their tendency to engage in single-electron transfer processes, typically giving rise to Wurtz-type homocoupling products (Scheme 9, B).³⁶ Finally, they readily undergo sulfoxide-magnesium exchange,³⁷ with complete retention of the stereochemical information, as illustrated in Scheme 9, C.^32b

Scheme 9: Examples of applications of Grignard reagents.

1.3.2 turbo-Grignard reagents

The latest breakthrough in the field of organomagnesium compounds is arguably Knochel’s discovery of the effect of LiCl on Grignard reagents.³⁸ The originally reported ⁱPrMgCl∙LiCl was named turbo-Grignard, and the term applies by extension to other LiCl complexed Grignard reagents. These compounds display an enhanced reactivity, in particular in halogen-magnesium exchange reactions, thus increasing the scope and availability of polyfunctionalized organomagnesium reagents.³⁹ Knochel attributed this enhancing effect of LiCl to its capacity to break multinuclear magnesium aggregates in favor of “ate” complexes (Scheme 10), which are more electron-rich and more nucleophilic. This was confirmed by Stalke and Koszinowski, who noted that while both Grignard reagents and their turbo- analogues showed similar structural behavior and equilibria, the amount of magnesate complex is significantly higher in the turbo-variant.

Scheme 10: Aggregation equilibrium of Grignard reagents with lithium chloride.

As mentioned earlier, turbo-Grignard reagents are effective at promoting the halogen-magnesium exchange reaction. This effect was first discovered in the metalation of electron-rich aryl halides, which were known to be poorly reactive in magnesium insertion. For instance, para-bromoanisole undergoes efficient exchange with ⁱPrMgCl∙LiCl to provide the corresponding new turbo-Grignard reagents in high yield (Scheme 11).³⁸

Scheme 11: Br-Mg exchange promoted by LiCl.

1.4 Magnesium amides

1.4.1 Hauser bases

The treatment of secondary amines, most commonly diisopropylamine (DIPA) or 2,2,6,6-tetramethylpiperidine (TMP), with a Grignard reagent yields magnesium amide species known as Hauser bases (Scheme 12).⁴⁰

Scheme 12: Preparation of Hauser bases DIPAMgCl and TMPMgCl.

Similarly to Grignard reagents, Hauser bases have challenged scientists with their complex and dynamic structural behavior. Stalke used DOSY NMR experiments to study the structure of DIPAMgCl in THF solutions (Scheme 13).⁴¹ Like Grignard reagents, Hauser bases exist as aggregates in a Schlenk equilibrium between homoleptic and heteroleptic species.

Concentration and temperature play a key role in the position of the equilibria: while the heteroleptic form is favored at high temperatures (above

‒20 °C), at lower temperatures the homoleptic species (ⁱPr2N)2Mg and MgCl2 dominate. Mulvey, Weatherstone and Hevia studied the TMP variant and were able to report crystal structures of the reagent in the solid state.⁴²

Scheme 13: Simplified representation of the Schlenk equilibria involving

iPr₂NMgCl.

Hauser bases were developed as an alternative to lithium amides, whose high reactivity is incompatible with sensitive electrophilic functional groups and often requires very low reaction temperatures. The most notable application of Hauser-bases is the ortho-metalation of carboxamides.⁴³ However, further investigation of their potential was hampered by their poor solubility in most organic solvents, a clear disadvantage in comparison with their lithium amide relatives.⁴¹

1.4.2 turbo-Hauser bases

After the discovery of the LiCl effect on Grignard reagents, Knochel developed analogous turbo-variants of the Hauser bases.⁴⁴ The preparation is straightforward using ⁱPrMgCl∙LiCl (see Scheme 12). The “turbo” prefix that is often used in the literature refers to their strongly increased basicity and reactivity. Stalke, Hevia and Mulvey studied the structures of both DIPAMgCl∙LiCl and TMPMgCl∙LiCl and found that the presence of LiCl has “an enormous impact” on the equilibria at play (Scheme 14).^{41, 45} It was found that LiCl shifts the Schlenk equilibrium towards the more reactive bimetallic species, both monomeric and dimeric, even at low temperatures (Scheme 14, left).

Scheme 14: Simplified structures of DIPAMgCl∙LiCl in solution.

The structural behavior of TMPMgCl∙LiCl is similar, but the increased steric hindrance provided by the TMP core further prevents the formation of dimeric species, favoring almost exclusively the bimetallic monomer.^45a Stalke suggests that since monomeric species are the most reactive, this explains why the TMP turbo-Hauser base is much more potent than its DIPA analogue in C-H metalation (Scheme 15, A). Indeed, the latter has been used only rarely since its first report, while the former has been applied for the direct magnesiation of a large variety of substrates, including the regioselective synthesis of heavily functionalized aryl and heteroaryl Grignard reagents (Scheme 15, B).^{44, 46}

Scheme 15: (A) Comparison of DIPAMgCl∙LiCl and TMPMgCl∙LiCl and (B) examples of application to direct magnesiation.

1.4.3 Organomagnesium amides

Compounds of the general formula R2NMgR’ (where R and R’ are organic ligands) are called organomagnesium amides, and can be prepared through various methods (Scheme 16).^26a They are generally synthesized from diorganomagnesium compounds by treatment with an amine, or with a metallic amide (Scheme 16, A). Alternatively, metallic amides can be combined with Grignard reagents to generate organomagnesium amides, releasing metal halides (Scheme 16, B). This method benefits from the availability of Grignard reagents, but their ethereal solvents have sometimes been correlated to a lower reactivity.^{26a, 47}

Scheme 16: Various methods exist for the preparation of organomagnesium amides.

Organomagnesium amides are structurally diverse, and many have been characterized crystallographically.^26a Most of these compounds are multimeric, although the degree of aggregation varies depending on the amine fragment (Scheme 17). For instance, the structure of DIPAMgBu is similar to the parent Hauser base (see section 1.4.1),^45b while the related dippNHMgEt forms a remarkable cyclic dodecamer featuring bridging amido and ethyl ligands.⁴⁸ In sharp contrast, the similar dippNHMgBu, when complexed with TMEDA, is mostly monomeric.^45b

Scheme 17: Organomagnesium amides display wildly different structural properties.

Organomagnesium amides have mostly been used in challenging metalation reactions.^26a The alkyl fragment acts as the base and releases the corresponding alkane (often as a gas) leading to an irreversible deprotonation, as shown in the example in Scheme 18.⁴⁹

Scheme 18: Regioselective deprotonation and functionalization of cyclopropane carboxamides using DIPAMgBu.

Organomagnesium amides have also been reported to react with carbonyl compounds (Scheme 19), to irreversibly generate enolates;⁵⁰ or to obtain alcohols through β-hydride transfer if the alkyl fragment is suitable.⁵¹ They can also react as mild nucleophiles to provide alkylation products (Scheme 19 and see section 4.1).⁵²

Scheme 19: Reactions of organomagnesium amides with carbonyl compounds.

1.5 Aims of the thesis

This thesis is aimed at developing new C‒C coupling methods using main- group organometallics. These widely available, inexpensive and relatively non-toxic reagents have been used for decades in a large variety of reactions as the most common source of nucleophilic carbon. Taking advantage of these solid groundings, emphasis is put on practical and economical reactions that can enable new advantageous retrosynthetic disconnections. In a broader sense, enabling shorter synthetic routes impacts the pace of discovery and the economic, human and environmental sustainability of synthesis.⁵³ With this research, we hope to facilitate the efficient and expedient synthesis of chemical building blocks, and ultimately more complex functional molecules such as pharmaceuticals and ligands.

In chapter 2, we report a synthesis of functionalized piperazine ligands through a [3+3] dimerization of azomethine ylides that is enabled by a radical mechanism. Chapter 3 describes a Pummerer-type reaction of sulfoxides and strong organomagnesium nucleophiles that is made possible by the use of a turbo-Hauser base. In the final two chapters, a similar strategy is reported as a solution for the challenging direct conversion of carboxylic acids to ketones with Grignard reagents.

2. Scalable Synthesis of Piperazines Enabled by Visible-Light Irradiation and Aluminum Organometallics

The contents of this chapter have been published in part in paper I.

2.1 Introduction

The deactivation of homogeneous catalysts is a common problem that limits important reactions and obscures mechanistic investigations.⁵⁴ In the context of C−H oxidation, the aggregation of the catalysts into stable, unreactive multinuclear clusters remains a challenge.⁵⁵ In principle, this oligomerization phenomenon could be addressed by the design of suitably engineered ligands that minimizes such aggregation. This issue has mainly been tackled by designing bulky ligands⁵⁶ that require multi-step syntheses. Janus-type ligands (who take their name from the Roman two-faced God Janus) offer new opportunities to modify the aggregation behavior of their complexes.^55b-

e, 57

However, they are mostly based on conjugated heterocyclic cores,^57a which are known to promote non-stereoretentive Fenton-type reactions. In this regard, piperazines are attractive, as they display a Janus-type topology together with a non-conjugated core. When functionalized with pyridine- based substituents, they are Janus-analogues of popular pyridylamine ligands that are commonly found in oxidation catalysts.57h, 57i, 58

However, the synthesis of densely functionalized piperazines is surprisingly challenging and currently requires tailor-made starting materials for multi-step C−N bond forming reactions, which suffer from stereocontrol issues.⁵⁹ A direct, straightforward synthesis of piperazines would therefore facilitate the exploration of these substrates in catalysis. From a retrosynthetic point of view, the most symmetrical disconnection across the two C−C bonds of the piperazine core would trace back to readily available imine starting materials 1 (Scheme 20). However, the derived azomethine ylides 2 are known to react in a [3+2] fashion to provide imidazolidine products 4.⁶⁰

Scheme 20: The native reactivity of azomethine ylides is [3+2] cycloaddition.

The desired [3+3] cycloaddition of aza-allyl anions is a thermally forbidden process which has been reported only rarely.⁶¹ Gibson, Xia and Li observed the formation of two piperazines while studying titanium complexes, in which the tridentate monoanionic backbone was necessary for the reaction to occur (Scheme 21, A).⁶¹ In depth studies of this phenomenon have been conducted by the Wolczianski group (B),⁶² who established the tendency of pyridine-functionalized 2-aza-allyl anion complexes of iron to undergo [3+3]

dimerization (Scheme 21, B). This represents a proof of principle, as these reactions are reversible and specific of a few substrates. The dimerization process is proposed to proceed via a radical mechanism.⁶²

Scheme 21: The formation of piperazines from imines bearing anionic substituents was observed by Gibson and Xia-Li.

Furthermore, piperazine formation had also been detected by Wang as a minor product upon treatment of a related amine substrate with trimethylaluminum (Scheme 22).⁶³ However, the authors explicitly stated that they could not reproduce this particular result.

Scheme 22: Traces of piperazines form by treatment of dipicolylamine with AlMe₃.

2.2 Aims of the project

The above-mentioned results suggest that a [3+3] cycloaddition of azomethine ylides may be possible through an SET process. We hypothesized that taking advantage of the photophysical properties of heterocyclic 2-aza-allyl-anions, visible light may be a suitable promoter to drive the synthesis of polyfunctionalized piperazines (Scheme 23).⁶⁴ We expected challenges to control the reversibility and he stereochemical outcome of this cycloaddition. Ultimately, we aimed to explore the potential of the resulting products as multidentate Janus-type ligands in challenging catalytic reactions.

Scheme 23: Synthetic strategy for the [3+3] cycloaddition of azomethine ylides.

2.3 Results and discussion

2.3.1 Reaction conditions

Wolczianski’s studies show that Lewis acids are instrumental to triggering the [3+3] dimerization process, while the main challenge is the reversibility of the process (see section 2.1).⁶² To overcome this, an ideal Lewis acid would need to be able to stabilize the piperazine bis-amide product (see Scheme 23) to prevent reversibility. In addition, it needs to be compatible with the strong bases needed to generate the azomethine ylide from the imine, and to be able to generate an electron-rich azomethine ylide suitable for single-electron transfer. Table 1 shows a variety of Lewis acids investigated for this purpose under visible light irradiation.⁶⁵ The potential involvement of a highly reactive diradical intermediate prompted the use of carefully deoxygenated solvents for our screening. Titanium reagents do not provide any product (entries 1, 2), possibly due to the absence of anionic substituents that had been succesful in a previous report (Scheme 21, A).⁶¹ Likewise, reagents based on zirconium, boron, silicon, zinc, copper or lithium are unable to promote the reaction (entries 3 – 7). Inspired by the results of Wang (with a different substrate, see Scheme 22),⁶³ we found that trimethylaluminum provides the desired piperazine 3a in 56% yield along with amine 5 as the main byproduct (entry 8). This amine likely stems from methyl transfer to the initial imine 1a from trimethylaluminum.

Table 1: Screening of Lewis acid reagents for the [3+3] dimerization of 1a.^[a]

Entry Reagent 3a^b (%) 5^b (%)

1 Ti(NEt)4 0 0

2 Ti(OⁱPr)₄ 0 0

3 ZrCl4 0 0

4 BF₃·OEt₂ 0 0

5 TMSOTf 0 0

6 ZnEt₂ 0 0

7 Me2CuLi 0 0

8 AlMe3 56 15

[a] Conditions: 1a (0.1 mmol) in indicated solvent (0.1 mL), light irradiation (from a 20 W cold fluorescent bulb) at 0 ⁰C; then indicated reagent (0.1 mmol), light irradiation at r.t. for 4 h. Toluene was degassed by three freeze-pump-thaw cycles. [b] Yields determined by ¹H NMR using 1,1,2,2-

Aluminum based Lewis acids were investigated further (Table 2).

Interestingly, arylaluminum reagents and aluminum-bases were found ineffective (entries 2 – 4). Furthermore, higher-alkylaluminum and aluminum hydrides lead to reduction of the imine substrate and did not provide any desired piperazine (entries 5, 6), highlighting the specificity of the trimethylaluminum promoter. Fortunately, the more coordinating solvent THF minimizes the formation of by-product 5 (entry 7), perhaps due to a decreased nucleophilicity of the organometallic, similarly to Grignard reagents. Other solvents were screened as well, although none proved superior to THF.⁶⁶ Among the various light sources investigated (entries 8, 9), white LEDs give the best yield and reproducibility (89 ± 3%, N = 9, entry 9). Additional control experiments show that while an extensive irradiation time only has a modest negative impact on the yield of 3a, the degassing of the solvent is crucial (entry 10, 11). Likewise, visible light irradiation is necessary (entry 12, see section 2.3.3).

Table 2: Optimization of the [3+3] dimerization of 1a with aluminum regents.^[a]

Entry Reagent Light source Solvent^b 3a^c (%) 5^c (%)

1 AlMe3 CFL^d Toluene 56 15

2 AlPh₃ CFL^d Toluene 0 0

3 Al(O^tBu)3 CFL^d Toluene 0 0

4 [Al(NR₂)₃]₂^e CFL^d Toluene 0 0

5 Al(ⁱBu)3 CFL^d Toluene 0^f 0

6 DIBAL-H CFL^d Toluene 0^f 0

7 AlMe3 CFL^d THF 60 2

8 AlMe₃ W^g THF 58 2

9 AlMe3 LED^h THF 89 3

10 AlMe₃ LED^h THF 74ⁱ 2

11 AlMe3 LED^h THF^j 11 23

12 AlMe₃ none THF 11 3

[a] Conditions: 1a (0.1 mmol) in indicated solvent (0.1 mL), light irradiation from the indicated light source at 0 ⁰C; then indicated reagent (0.1 mmol), light irradiation at r.t. for 4 h. [b] The solvent was degassed by three freeze-pump-thaw cycles. [c] Yields determined by ¹H NMR using 1,1,2,2- tetrachloroethane as internal standard. [d] Standard household light bulb 20 W. [e] NR2 = pyrrolidin-1-yl.

[f] Imine reduction observed as the major product [g] Tungsten filament bulb 100 W. [h] 14.4 W cold white LED strip. [i] 24 h reaction time. [j] Solvent not degassed. CFL = compact fluorescent light, LED =

2.3.2 Substrate scope

The optimal conditions were applied to a set of symmetrical imines, observing the formation of piperazines 3 as a single all-equatorial diastereoisomer (Scheme 24). Alkyl substituents are tolerated on all four available positions of the pyridine ring, providing good to excellent yields (3a-d), with the notable exception of 3e, which was obtained in only 37%

yield. This is likely due to steric hindrance caused by the methyl substituent on position 6, which disfavors the interaction with the aluminum center within the azomethine ylide complex. All products are conveniently obtained by recrystallization, and suitable for gram-scale synthesis, as illustrated with 3a.

Scheme 24: Scope of alkyl-pyridines-substituted imines. Conditions: 1 (0.2 mmol) in THF (0.2 mL), LED irradiation at 0 ⁰C; then AlMe₃ (0.2 mmol), LED irradiation at r.t. for 4 h. Isolated yields after purification by recrystallization. [a] Reaction time 20 h.

Imines containing non-alkyl-substituted pyridine rings also efficiently engage in this [3+3] cycloaddition (Scheme 25). Pyridines bearing electron- withdrawing (3f, g) and electron-donating (3h) groups are tolerated with similarly good efficiency. Likewise, other heterocyclic scaffolds like the electron-deficient pyrazine and the electron-rich imidazole are compatible with these conditions, providing the corresponding piperazines 3i, j.

Unsymmetrical imines also provide piperazines in excellent yields, and in

of the regioisomer formed, with the identical groups in a 1,2-disposition, was confirmed by X-ray diffraction analysis. Remarkably, in the similar titanium system that was successful in [3+3] cycloadditions with anionic substituents, the regioselectivity observed was opposite⁶¹ (i.e., with the identical groups in 1,4-disposition, Scheme 21, A), which suggests that different mechanisms are operating in these two systems. Interestingly, 3ia was obtained as a 1.6:1 mixture of regioisomers, which leads us to speculate that imines featuring a strongly electron-donating and a strongly electron-withdrawing group (“push-pull”) may be required to obtain a high regioselectivity.

Scheme 25: Substrate scope with variously substituted N-heterocycles. For detailed reaction conditions, see Scheme 24. Isolated yields after purification by recrystallization. [a] Reaction time 16 h.

2.3.3 Mechanistic studies

Control experiments were run to investigate the essential role of the aluminum reagent in this process, and are summarized in Scheme 26.

Following the reaction of AlMe3 and imine 1a by ¹H NMR, it is possible to witness the formation of the azomethine ylide intermediate 2a as well as to observe the formation of methane as a by-product, which demonstrates the dual role of AlMe3 as a base and a [Me2Al]⁺ source in this process. In the absence of visible-light, there is only margnial conversion. Moreover, piperazine 3a can be prepared via treatment of the Li(smif) 2b with AlMe2Cl; however, irradiation of 2b in the absence of the aluminum reagent does not provide any piperazine, thus demonstrating the essential role of both visible-light and the organoaluminum handle. We currently rationalize the reactivity observed with trimethylaluminum over other organoaluminum reagents based on the stronger electron-donating ability of the methyl ligands. We speculate that this effect increases the energy of the HOMO of the CNC backbone of 2a, enabling the single-electron-transfer to the pyridine π* orbital to occur.

Scheme 26: Although the azomethine ylide can be generated through an alternative pathway, both light and [AlMe₂]⁺ are essential to produce piperazine 3a.

The effect of alternating periods of visible-light irradiation and darkness clearly shows the correlation between illumination and formation of 3a (Figure 1). We observe a slight conversion to 3a in the absence of light that stalls over time. Although we reason that this is a photo-mediated reaction, our current data cannot rule out a photo-initiated process with low chain- carrying capacity.⁶⁷

Figure 1: The formation of 3a is clearly accelerated by visible-light irradiation.

Experiments thermostatized at 0 ⁰C. Yields determined by ¹H NMR using 1,1,2,2- tetrachloroethane as internal standard. Grey shadow indicates the absence of light in the light on/off experiment.

2.3.4 Application of the products in catalysis

To investigate the potential of the products as ligands in iron catalyzed C−H oxidation, 3a was N-alkylated to provide Janus-type ligand 6, which we termed PiPy6 (Scheme 27, A). The corresponding dinuclear iron catalyst 7 was then prepared using iron triflate. We hypothesized that the highly congested environment provided by the PiPy6 ligand would hinder the dimerization of the catalyst. While X-ray analysis shows the presence of two water molecules, the dinuclear complex is monomeric. This is further confirmed by HRMS analysis, in which only the monomer was observed even at high water concentrations. In contrast, analogues of this catalyst bearing the benchmark TPA ligand 8 readily form oxo-bridged dimers in presence of water, which are catalytically inactive (Scheme 27, B).55a-e, 57g, 58e, 68

0 20 40 60 80 100

0 1 2 3 4 5

3a yield (%)

time (h)

background thermal reaction (light off) light on/off (see grey zones)

standard conditions (light on)

Scheme 27: (A) Synthesis of the dinuclear iron catalyst 7 and (B) Formation of a dinuclear iron complex 9 in the presence of water.^55a

The performance of dinuclear monomeric catalyst 7 was compared with that of the TPA analogue 9 on the oxidation of cis-dimethylcyclohexane (Table 3).^{58d, 58e} It was found that for the same amounts of iron, the dinuclear system performed better, further supporting the hypothesis that the congested ligand environment limits the oligomerization deactivation pathway observed with TPA.

Table 3: Comparison of the catalytic activity of FeTPA and Fe₂PiPy₆.^[a]

Entry Catalyst Catalyst loading 11^b (%)

1 FeTPA (9) 5 mol% 6.8

2 Fe2PiPy6 (7) 2.5 mol% 10.9

[a] Conditions: 10 (0.2 mmol) in MeCN (1.0 mL) and the indicated catalyst (0.05 – 0.1 M in MeCN) added to dry flask at r.t. under Ar; then acetic acid (0.1 mmol); then H2O2 (0.26 mmol, 1.5 M in MeCN) over 6 min via syringe-pump; then r.t., 10 min. [b] Yield determined by GC-MS using 1,2- dichlorobenzene as internal standard, average of three experiments.

2.4 Conclusion and outlook

In summary, an efficient and scalable procedure for the synthesis of densely functionalized piperazines has been developed. This protocol relies on an unusual combination of visible light irradiation and aluminum organometallics that allows the rerouting of the native [3+2] reactivity of azomethine ylides into a [3+3] cycloaddition. The heterocyclic ring of the imine starting materials provides an acceptor for the intraligand charge transfer, and can be decorated in all positions with electron-rich and electron-poor functional groups alike. Furthermore, unsymmetrical imines provide piperazine products with an unexpected, complete regioselectivity, further showcasing the potential of this strategy for the synthesis of designed ligands, whose potential in catalysis has been preliminarily demonstrated.

Additional studies are required to obtain a deeper understanding of the mechanistic features of the reaction, in particular the role of the light, the specificity of trimethylaluminum and the origin of the unusual regioselectivity observed in unsymmetrical substrates.

3. Intermolecular Pummerer Coupling with Carbon Nucleophiles in Non-Electrophilic Media

The contents of this chapter have been published in part in papers II and III.

3.1 Introduction

3.1.1 Thioethers

The natural abundance and the unique properties of sulfur have made it a central element in Nature as well as in chemistry.⁶⁹ The thioether moiety is thus found in a large number of natural products,^69a-c pharmaceuticals,^69b-f and ligands(Figure 2).^{69g, 69h} For instance, methionine is an essential amino acid, biotin has found important applications in protein tagging,^{69i, 69j} and volatile thioethers such as tropathiane are common ingredients in the fragrance industry.^{69i, 70} The widely used β-lactamic antibiotics penicillins and cephalosporins also contain the thioether moiety,^{69b, 71} and synthetic thioether-based drugs such as benzothiazepine derivatives are common as well.^69f

Figure 2: Thioethers are common in both natural and synthetic products.